(Doe) Parties: When Can Politicians Combine Clientelistic and Programmatic Appeals?

“Do Everything” (DoE) Parties: When Can Politicians Combine Clientelistic and Programmatic Appeals?

Matthew Singer University of Connecticut

Herbert Kitschelt Duke University

Paper Sketch prepared for presentation at the Workshop on Democratic Accountability Strategies. Duke University. Durham, May 18-19, 2011.

This paper is based on a dataset collected under the auspices of the project “Political Accountability in Democratic Party Competition and Economic Governance,” implemented by a political science research group at Duke University (Principal Investigator: Herbert Kitschelt). Funding for the data collection was provided by the World Bank, Duke University, and the Chilean Science Foundation (research grant directed by Juan Pablo Luna and David Altman, Catholic University of Chile). Data analysis and conclusions of this paper are the sole responsibility of its author(s).

Introduction

Models of how parties choose their linkage strategies usually posit that parties face a tradeoff between programmatic competition and clientelism (e.g. Dixit and Londregan 1996, Kitschelt 2000, Cox and McCubbins 2001, Keefer 2007, Lyne 2008). This theoretical modeling assumption has considerable plausibility, but direct evidence relies empirically on individual case studies or small-N comparisons. Yet a few studies have pointed out that parties may engage in product diversification and appeal to different constituencies with different linkage efforts (Magaloni et al 2009). The data for a large-scale global test of the assumption with observations across a broad spectrum of political parties have not been available until now.

The Democratic Accountability project expert survey on political parties for the first time delivers the information to test the trade-off hypothesis on a global scale. Initial data analysis indeed suggests that the trade-off hypothesis is borne out. But as is the case almost invariably with large-N empirical investigations, the observations yield somewhat fuzzier and messier conclusions than theorists might have hoped for. Indeed, among the 506 parties in 88 countries included in the data collection there are rather substantial numbers of parties that defy the tradeoff hypothesis. At one extreme, there are parties that are strenuously making an effort to provide a variety of linkage mechanisms with their electoral constituencies. These “Do Everything” (DoE) parties display not only a capacity to reach out to electoral constituencies through vigorous clientelistic efforts to target material benefits to individuals and small groups of voters, but also a penchant to develop rather sharp programmatic policy contours, if elected to form a government. At the other extreme, there is a large subset of parties that, in the judgment of experts, are neither good at clientelistic outreach nor at programmatic appeal.

What makes it possible for ‘Good for Nothing’ and “Do Everything” parties to appear and even rise to prominence? Among the two configurations of linkage outreach that defy the tradeoff hypothesis, DoE parties pose the greater and more interesting theoretical challenge. ‘Good for Nothing’ parties, by contrasts, tend to be electoral underperformers with small followings, sometimes even outright fringe parties. These parties typically tend to be of too recent origin to have been able to make a serious investment in party organization and linkage strategies. Or they are waning parties that have lost their way or found niches in which they play a small role in a highly fragmented system. In addition to these systematic effects, of course, a lot of parties identified as Do Nothing may end up in this category by virtue of measurement error only.

In this paper sketch, we will thus take up the argument primarily for the more complicated and rewarding object of analysis, that of DoE parties. How can they push the envelope of democratic linkage building in so many different directions and sustain highly diversified electoral appeals. In the first section, we discuss the conditions that theoretically may give rise to DoE parties and the organizational and network choices that might sustain them. In the second section, we identify DoE parties in the Democratic Accountability Dataset and examine a range of complementary features that distinguish DoE parties from the rest of the field of political parties. We are here concerned with the organization of DoE parties and their use of social and associational networks. In the third and most tentative section, we probe into the

2 origins of DoEs. We realize that much more headway needs to be made than we will supply to come up with an elegantly simple, yet empirically powerful account of DoE parties.

1. Rise and Persistence of DoE Parties. Causal Conditions and Complementarities

Effective clientelistic and programmatic mobilization involve distinct patterns of electoral demand, features of party organization, and governance structures to allocate benefits to partisan supporters. Trying to combine these under the umbrella of a single party brand imposes contradictory demands on party activists and leaders. These strains make it unlikely that parties actually can combine clientelistic and programmatic accountability. More specifically, there are at least four features where clientelistic and programmatic politics get in each other’s way:

1. Political Governance: It takes different kinds of laws and administrative governance structures to use the policy process to target policies through clientelist channels compared to programmatic ones. Clientelism relies on vague legislation with lots of discretion concerning resource allocation at the operational level of disbursement. Programmatic politics works through highly codified, universalistic conditional programs with clearly specified eligibility criteria. It is thus likely to be difficult to run both types of governance in the same party regime.

2. Distribution of Electoral Demand: The electorates to whom clientelism appeals may not only be different from those that prefer programmatic party competition, but directly at loggerheads with each other. Members of the constituencies with whom programmatic appeals resonate often despise voters craving for clientelistic appeals. This applies especially to more educated “middle class” voting blocs. “Progressive” movements against clientelism typically originate with urban middle class voters.

3. Party Resource Allocation: Clientelistic politics requires access to extraordinary quantities of tangible resources—either through private donors (vote-poor, resource-rich constituencies…) and/or through public funds made available after the capture of executive office. Many parties simply have no access to such resources (Shefter 1977). Even when parties have substantial resources, they may have to decide carefully whether to deploy them in clientelistic or programmatic fashion. Resource constraints may force parties to abandon clientelistic bids. At the margin, parties will often have to choose whether to invest in clientelistic or programmatic politics.

4. Party Organization: Clientelism and programmatic competition also place different demands on party organizational structures. Clientelism works best in party organizations based on encompassing informal networks of notables and external interest associations and with a highly centralized political staff that can funnel resources into the organization and on to external clients in an opaque, clandestine fashion. Programmatic parties, by contrast, can solve their credibility problems better by more or less extensive formal organization, combined with a leadership structure that imposes compliance with formal

transparency in the handling of resources and the recruitment of personnel and a modicum of institutionalized veto players as checks on the parties’ executive leaderships.

Now, just because parties face tradeoffs in jointly pursuing clientelist and programmatic strategies at the same time does not imply that parties will not attempt to engage in both behaviors. Yet we generally expect that doing both to the fullest extent will be difficult. These mechanisms do suggest, however, ways in which parties may be able to combine both strategies. Parties may have access to specific historical legacies, economic groups within society, or extreme resources that allow them to overcome the constraints these tradeoffs assume exist. Thus we suggest a few counterfactuals that should make DoEs viable in defiance of the tradeoff postulate. The first two mechanisms (governance, electorates) speak to the causal origins of DoE parties, the latter two (finances and organizational structures) to the complementarities involved in setting up clientelistic and programmatic linkages.

Causal Trajectories: Governance and Electorates

The critical condition that may enable parties to compatibilize clientelistic and programmatic appeals has to do with economic development and the role of central state intervention. Let us use a simple stylized development model as the foil for our argument, distinguishing poor, medium income, and high income developed economies. In poor economies, economic activity is primarily organized at the local level and political players ask for few large club goods and collective goods only a high-capacity national state can provide. In this setting, parties are likely to coordinate around clientelistic accountability mechanisms, while public administration is shot through with patronage, capture by localized rent-seekers and more generally a pattern of governance in which idiosyncratic and personalized networks reign supreme. In wealthy, advanced economies, at the other extreme, local economies are fully tied into national and global economies. While there will still be plenty of particularistic rent seekers, they have a relatively weaker hand than ever before, with electorates, parties and governments being more concerned about the allocation of large-scale club goods and collective goods. Here political governance works more through generalized rules and impersonal institutionalized procedures than informal networks, rendering it implausible that politicians can expect much success with efforts to develop clientelistic networks.

This leaves as critical window an intermediate development level, where local production systems erode and are being partially displaced by national and international markets and corporations. The state may take a particularly active role in compensating the losers of the transition process as well as pushing the winners of the process forward. In political-economic governance terms, this intermediate situation has been organized variably either in systems of import substituting industrialization (ISI) or export-oriented industrialization (EOI) (Haggard 1990). Without discussing the rather different economic performance consequences of each governance variant, they share in common the mobilization of national state capacity to provide collective and large-scale club goods to the economy, while at the same time intervening in the allocation of highly specific resources in local settings.

In a very roundabout way, we therefore wish to hypothesize that political parties have the greatest chances to develop DoE appeals, if they are embedded in settings situated at this intermediate stage of economic development. On the one hand, they are facing electorates with whom models of national economic development model resonate and that are concerned about the large-scale distributive consequences of alternative models for the relative performance of economic factors (capital, labor, land) and sectors (export oriented/exposed versus domestic; agriculture and manufacturing) of the economy. On the other, they face large electorates that demand compensatory side-payments for the erosion of local economies. Whereas in advanced industrial democracies this compensation is for structural change is organized through the welfare state (Iversen and Cusack 2000), developing countries at intermediate levels may not have the resources and the governance capabilities to deliver such social policy regimes. Political parties will therefore more likely resort to clientelistic pay-off strategies, such as public works in the countryside, large state-run enterprises, or employment through over-staffed public sector agencies (with the post office as an example).

What makes urban middle strata acquiescent to such arrangements and preempts “progressive” movements until a later date is precisely the state-led nature of the political economy. Under such governance regimes, large portions of the emerging urban middle class are themselves dependent in their employment on economic state intervention. They will tolerate a combination of (programmatic) development model with (clientelistic) side-payments, as long as they see themselves as beneficiaries of this Janus-faced set-up.

The simple prediction that follows from this is that DoE parties should be most successful in middle income countries. They should be both most numerous in such countries, relative to other configurations of parties, and most electorally successful. As a more complicated hypothesis, we expect some association of DoE parties with state-interventionist ISI or EOI regimes. But this association may appear in different configurations. Let us distinguish two cases. At the lower bound of middle income development, where ISI/EOI regimes have never existed before, but where countries enter the “window” of opportunities for state-led economic development, DoE parties are likely to be new challengers who project an aggressive, redistributive, populist economic program of development, coupled with clientelistic strategies to compensate the losers. At the upper bound of middle income development, when ISI and EOI strategies begin to deliver highly volatile and crisis-ridden economic performance, it will be mostly older, established political parties that sustained the ISI/EOI effort in a democratic setting. Where ISI/EOI was primarily pushed by authoritarian systems of rule, agents of such regimes will assemble in DoE parties and use their network linkages to build both a clientelistic following as well as a following based on a redistributive-populist political appeal. Thus, some parties coming out of the ISI/EOI regimes should be able to “segregate” the support of different constituencies, but combine them under the same umbrella (e.g. Levitsky 2003 and 2007), especially if they can loosen resource constraints on party-led clientelistic disbursements by capturing the government.

A challenge is testing this story (that we have not yet overcome) is that countries at an intermediate level of development should also be at a cross-roads of dominant linkage types. Clientelism is becoming increasingly untenable but parties have not yet development the reputations or electoral bases necessary for pure programmatic competition. Thus we expect that

5 intermediate income countries will not only have DoE parties but also parties that make a less successful mix of clientelist and programmatic linkages.

Beyond the general political economy story, DoEs may be supported by specific conditions of politically mobilized ethnic pluralism, particularly if it is combined with a situation of group status based political domination or on economic inequality between ethnic groups. From the vantage point of individual citizens, group-based asymmetries turn ethnic groups into “communities of fate,” in which the economic and socio-political payoffs of each member in the political economy derive not simply from her individual endowments, effort or luck, but are tied to the opportunities of all members of the group as a whole. DoE parties may come in here in one of two guises. The first variant is a party that exclusively represents a disadvantaged ethnocultural group. It will struggle for an economic program of general redistribution to the poor, while also seeking clientelistic benefits for all members of its own group, regardless of individual standing. An example would be the native Indian party Pachakutik in Ecuador. The second variant is a large party encompassing various ethnicities that ventures to organize a consociational political compact. It might also opt for (some) economic redistribution based on individual income position, but otherwise rely on clientelistic benefits disbursed contingent upon group membership and political support for the party.

Functional Complementarities: Party Organization and Finances

Complementarities of DoE strategies are all those activities and investments politicians make more or less simultaneously with deciding over a set of linkage strategies. When we say “set” here, we might include also linkage strategies other than clientelism and programmatic politics. Complementarities furthermore concern the organizational form of the party and its network embeddedness in society and political economy. The size of the party and its government status may also be critical complementarities, particularly for the process of acquiring and distribution resources. In general, these strategic choices are aimed at acquiring resources (financial, organizational, and legitimizing) that allow parties to attend to the organizational imperatives of offering two forms of electoral appeal.

First, the demands of clientelist distribution and monitoring suggest that DoE parties will have particularly extensive formal party organizations. The large formal organization is complemented by a broad spectrum of network ties both into informal local connections to local notables as well as more explicit interfaces with interest associations configured around a variety of causes.

Following Kitschelt and Kselman (2010), in terms of political leadership, we would expect DoE parties to incorporate fewer features of checks and balances and to permit more complete control over relevant decisions—political careers and nominations for electoral office, choice of external party strategies—by a small group of leaders than would be common in purely programmatic parties. The underlying logic of intra-party authority concerns the pre- commitments parties must be able to make to programmatically motivated, rational voters who want to support parties with policy proposals close to their own personal ideal points. While such voters do not want to see the party at the mercy of small gangs of ideological party activists, they also would like to see limits imposed on the autonomy and discretion accorded to the party’s

6 leadership. None of these constraints govern clientelistic parties, where voters and activists are strictly interested in the flow of material selective incentives. Here constituencies and activists could care less about the nature of party authority, as long as the leadership is able to deliver the benefits on time.

At the same time, DoE parties have incentives to maintain their ideological consistency and strength. Part of this can be done by staking out strong policy positions outside the electoral center-positing a unique vision for the center. They also have incentives to build affective attachments to the party and cultivate an expressive linkage based on party identification. Finally, given the need for resources and the advantages of government status, DoE parties may feature valence appeals and try to highlight their unique capacities to govern competently as a linkage appeal.

Maintaining these structures requires substantial financial inputs, as does the costs involved in running a clientelist campaign. Electoral office provides an important access point to extract and distribute resources to clientelistic networks. This should make it more likely that electorally larger, successful parties display DoE features, without judging here whether the causal arrow runs from DoE appeals to electoral growth or the other way round. DoEs are also likely to be frequent governing parties. Large DoE parties may also be characterized by exceptionally murky finances to sustain the voracious demands of servicing a broad and complex array of constituencies.

Finally, because they put a premium on centralized leadership and an opaque flow of financial resources, it is likely that leadership personalities play an exceptional role in DoE parties. They should exhibit a fairly strong dosage of “charismatic” authority. This authority may also provide an alternative way of smoothing conflict when clientelist and charismatic goals collide.

2. Identifying “Do Everything” Parties Party Strategies and Complementary Features

The Democratic Accountability project has identified clientelistic and programmatic party appeals in a variety of ways. The theoretically most appropriate operationalizations for our purpose are the measures of politicians “efforts” that derive from the detailed questions of the expert survey. On the side of clientelism, this is the summary index of partisan effort (B15) that results from the addition of experts’ party scores on their propensity to produce targeted benefits for constituency in form of (1) gifts, (2) privileged access to social policy entitlements, (3) jobs, (4) procurement contracts and (5) favorable regulatory rulings. On the side of programmatic appeals, it is a summary measure of parties programmatic stances on four policy issues in each country that takes into account with how much internal unity a party voices positions on these issues (Cohension), how much importance for their own strategy parties attribute to the issues (Salience) and how much parties set themselves apart from each other on the issues (polarization). The resulting index is a product of the party scores on these three dimensions over four issues (CoSalPo_4).

Constructing this measure for 506 parties, what is the relationship between clientelistic partisan effort (B15) and parties’ programmatic appeals (CoSalPo_4)? Figure 1 provides the answer. There is indeed a fairly strong negative correlation (r = -.44), but a quick inspection of the scattergram reveals some spectacular outliers. A larger number of cases in the upper right quadrant of the figure show elevated levels of programmatic appeals despite displaying rather intense clientelistic efforts as well. These parties are in the “Do Everything” region. In the “Do Nothing” region of the figure the extreme places in the space are empty, but there are substantial numbers of observations with parties showing both feeble programmatic elaboration as well as weak clientelistic efforts.

(Figure 1 about here)

For the purposes of a comparison of parties pursuing different packages of clientelistic and programmatic linkage strategies, there are five general ways in which clientelist and programmatic linkages can be combined. Figure 2 lays out the conceptual distribution of cases. The first three reflect positions along a single clientelism-programmatic axis that suggests a tradeoff between clientelism and program:

Pure Program: Clear programmatic positioning, low levels of clientelist distribution Pure Clientelism: High levels of clientelist distribution, unclear programmatic positions Intermediate Mix of Strategies: Moderate levels of clientelist exchange and programmatic clarity, with neither to full effect

The other two strategy profiles lie off that axis:

Do Everything: High levels of clientelist distribution while maintaining high levels of programmatic clarity Do Nothing: Low levels of programmatic clarity and little distribution of clientelist benefits.

(Figure 2 about here)

A challenge in classifying parties into these five rough types arises because our measures of clientelist and programmatic competition do not have a natural cut points of medium, high, and low levels of each strategy. In this chapter, we categorize cases according to their relationship to the mean level of clientelism (12.12) and programmatic competition (0.2379). Specifically, we begin by classifying a party as being high (low) on a strategy dimension, if it is above (below) the mean value on that dimension (Figure 3). We then classify parties as being in the intermediate camp if they are in an ellipse that has a radius of one standard deviation away from the joint mean.1 In general, most parties can be classified as being either purely clientelist,

1 In making this choice, we considered three questions. One question is how to define the center of the distribution. The mean value may be affected by outliers while the median value imposes a symmetry in the data (i.e. the number of clientelist parties must equal the number of programmatic ones). A related question is where we should divide intermediate values of each strategy from high values. The third question concerns the shape of the intermediate region; should it be a rectangle as in figure 2 or a circle or some other shape that classifies parties according to their overall divergence from the central intermediate strategy profile. However, in alternative empirical specifications 8 purely programmatic, or in the intermediate category. Only 21 percent of cases are categorized as either Do Everything or Do Nothing parties, with Do Everything parties being especially rare. The relative paucity of Do Everything parties is what we generally expected-most parties should not be able to successfully use clientelism and program.

(Figure 3 about here)

Table 1 breaks down the distribution of party type by region and provides some description of their basic characteristics. The distribution of party types in Table 1 has a distinct regional component. Do Everything parties are non-existent in Sub Saharan Africa (although there is one in Egypt) and are rare in the Middle East and Asia. In these two regions, pure clientelist parties abound, although there are large number of parties that attempt to mix clientelist and programmatic appeals. These parties, however, make not sufficient effort to cultivate either linkage strategy to fit our Do Everything categorization. DoE parties are also rare in advanced capitalist economies, which are dominated by purely programmatic parties along with a surprising number of parties that do not have a distinct ideological profile or clientelist network.

(Table 1 about here)

The following 44 parties, broken down by region, fit our empirical conception of Do Everything parties. They exist in 27 countries.

Advanced Capitalist Economies: Belgium (Socialist Party), Greece (New Democracy, Panhellenic Socialist Movement), Italy (Forward Italy), Japan (Liberal Democratic Party), Portugal (Social Democratic Party)

Post-Communist: Albania (Democratic Party of Albania, Socialist Party of Albania), Croatia (Croatian Democratic Union, Croatian Peasant Party), Estonia (Centre Party of Estonia), Macedonia (New Social Democratic Party, The Internal Macedonian Revolutionary Organization – Democratic Party for Macedonian National Unity), Moldova (Communist Party of Moldova), Romania (Social Democrat Party), Russia (United Russia), Serbia (Democratic Party, G17+, New Serbia, Serbian Radical Party, Socialist Party of Serbia), Slovakia (Direction - Social Democracy)

Latin America: Bolivia (Movement for Socialism), Colombia (Colombian Conservative Party, Social National Unity Party), Ecuador (Democratic Left, Ecuadorian Roldosist Party, Institutional Renewal Party of National Action, January 21Patriotic Society Party, Movement Plurinational Pachakutik Unity - New Country, Social Christian Party), Guatemala (National Unity of Hope), Mexico (Institutional Revolutionary Party, National Action Party, Party of the Democratic Revolution), Panama (Panameñista Party), Paraguay (The Authentic Radical Liberal Party)

we explore in the appendix, we have also used the median as a measure of central tendency, used the 25th/75th percentiles as dividing points, and changed the shape of the intermediate region into a rectangle. Although the specific classification of cases differs across the specifications, the Cramer’s V between these classification schemes is in each case greater than 0.80. The empirical conclusions reached in the empirical analysis below … 9

Middle East/Asia: Egypt (National Democratic Party), Israel (Shas), Malaysia (Malaysian Chinese Association, United Malays National Organization), Pakistan (Pakistan Peoples Party), South Korea (The Grand National Party), Turkey (Nationalist Movement Party)

Do everything parties have the largest presence in Mexico (where the three major parties that combined to win over 90% of the seats in the last two legislative elections are all classified as Do Everything parties) and Greece (the two do-everything parties combine to win 89 percent of the seats).

Another way to consider the distribution of party types is to consider what kind of party is the largest party in each country. In Table 2, we divide countries according to the strategy used by the party that had the largest average representation in the last two legislative elections. The first two columns confirm the global importance of clientelism; most of the countries in our 88 country sample are dominated by a party that has high levels of clientelism. The 18 countries where high levels of clientelism are combined with programmatic positioning are again heavily drawn from Eastern Europe and Latin America.

(Table 2 about here)

We suggested that Do everything parties should tend to have a general institutional and organizational type that complements their specific mix of policy appeals. Thus before we start dwelling on the trajectories that may allow politicians to construct DoEs, let us consider the features that complement the parties’ linkage efforts. Each of the following tables is laid out in a particular pattern that should facilitate comparison. Most of the time, we expect extreme cases to be pure clientelistic parties OR DoE parties, at one extreme, and pure programmatic, or in fact Do Nothing, parties on the other. Hence the tables are organized in that order.

The first table simply supplies some sense of the parties’ size and access to government. Is it the case that large parties entrenched in the political executive have greater opportunities to pursue DoE linkage strategies? Table 3 shows that this in fact is the case. In general, parties that use clientelism are more likely to be large and in government than are those that use program; being in government increases the likelihood that a party has materialist benefits to distribute while parties may use materialist handouts to bridge ideological divisions that would otherwise divide them. Yet Do Everything parties are especially likely to be large. The majority of DoE parties are in government and their average legislative representation is larger than among any other group. More than 70 percent of DoE parties are either the largest or second largest party in their country, compared to only 51 percent of clientelist parties and less than 25 percent of the other types.

(Table 3 about here)

Do everything parties were expected to follow the basic organizational structure of clientelist parties by investing in extensive local party networks. They are more likely to have local offices than any other party type (Table 4). Do everything parties compliment these local offices by also maintaining ties to local intermediaries (e.g. neighborhood leaders, local notables, religious leaders) who operate in local constituencies on the parties’ behalf and work at maintaining contact with large groups of voters, organizing electoral support and voter turnout,

10 and distributing party resources. In this they are a slight step ahead of clientelist and intermediate mix parties. Clientelist parties in contrast are slightly more likely than Do Everything parties to have linkages with organized civil society groups and to use those groups to distribute materialist handouts. Do everything and clientelist parties thus seem to differ in their model of distribution and monitoring, with pure clientelist organizations more likely to delegate these activities to other organized groups in society while Do Everything parties either subcontract to intermediaries or keep it in house through their network of local party officials.

(Table 4 about here)

Intermediate parties face the same organizational challenges as Do Everything parties, but are less successful in building the necessary networks. They are less likely to have ties to civil society organizations than clientelist and Do Everything parties. They try to make up for this missing network by having the second highest level of local offices, but they trail Do Everything parties in this regard. The gap between the two types of mixed strategy parties becomes larger when looking at their ties to local notables.

As a result of these efforts to build a local distribution network, respondents in the expert survey consider DoE parties to be just as effective in their clientelist appeals as pure clientelist parties are (Table 5). Do Everything parties are just as likely to monitor voters and sanction defectors as pure clientelist parties are. Clientelist and Do Everything parties are successful because they can engage in monitoring and distribution through intermediaries, but the mechanisms through which this is accomplished differ. Intermediate parties, in contrast, are less effective in their use of clientelism than are pure clientelist or DoE parties. In particular, intermediate parties are less able to monitor their voters after they receive benefits from their party to ensure that they vote for them than clientelist parties are and they are also less likely to sanction voters who break their promises.2 Thus as they do not fully commit to clientelist mobilization, these parties are also less willing to violate the secrecy of the ballot box. Do Everything parties, in contrast, have found a way to overcome the organizational difficulties that intermediate parties face.

(Table 5 about here)

While Do Everything parties maintain local offices, they do not decentralize political decision making (Table 6). Do everything parties, for example, are more likely to centralize control over nominations than any other party type and they are also the least likely to give party members a role in nominating candidates through a primary or caucus system. Thus Do Everything parties are not engaging in bottom up policy discussions despite their extensive local party organization. Local offices are less likely to be instruments of interest aggregation than sites for dissemination of information and material goods. Intermediate mix and pure clientelist parties are also relatively less likely to decentralize nomination authority.

2 In the web appendix we test if monitoring and sanctioning are lower among intermediate parties than among clientelist and do everything parties when controlling for their size and the level of development in the country and these differences are significant at the 0.05 level. 11

In building these networks, Do Everything and Clientelist parties are likely to require financial resources. A portion of these resources they gain illegally through violating campaign finance regulations (Table 6, also Singer this volume). Yet clientelist parties break campaign finance rules with greater frequency than Do Everything parties. One possible explanation for this difference is that the programmatic components of the DoE strategy requires parties maintain clean hands. An alternative possibility is that DoE parties are relying on advantages that come because they are substantially more likely to be in government and have a large legislative delegation than is any other kind of party; large parties are less likely to break campaign finance laws than are small parties (Singer, this volume Table 4).

Do Everything parties also complement their use of programmatic and clientelist appeals with a reliance on other political legitimization strategies. The expert survey also asked respondents to rate parties according to their efforts to use appeals based on a charismatic leader, to build party identification within the electorate, and to focus on valence issues that emphasize the party’s governing competence to handle valence issues. Do everything parties thus do actually try do seem to try to do everything they possible can to form bonds with voters. The highest levels of all these indicators of party linkage effort are given to Do Everything parties. Pure clientelist parties are relatively more likely to use these alternative linkage strategies (especially charismatic leaders) than are programmatic parties (Table 7). Intermediate mix parties have intermediate levels of party identification and valence linkage but have lower levels of charismatic leadership than do clientelist and Do Everything parties.

(Table 7 about here)

Yet while Do Everything parties diverge from programmatic parties in their use of multiple forms of linkage and their organizational principles, one way in which they are similar is in their tendency to take political positions outside the ideological center. For example, in Table 7 we categorize parties according to whether our experts placed them in the middle third of the ideological spectrum. Only 28 percent of pure programmatic parties are in this region- programmatic parties are more likely to take an extreme ideological position than a centrist one. Clientelist parties, in contrast, are more likely to be closer to the center. Do Everything parties are more likely to be centrist than programmatic parties are but less centrist than clientelist parties or the other mixed types.

3. Development of Do Everything Parties

The first argument we expressed was that it is middle income countries where DoEs should be particularly prominent, as an urban middle class develops and begins to demand collective goods and as economic development necessitates more investments in general infrastructure and collective goods. At the same time, state capacities are rising and delivering resources that can be deployed by political parties. The regional distribution of DoE countries highlighted above speaks to this problematic. DoEs are just about nowhere to be found in the poorest regions on earth (Sub-Saharan Africa, parts of South Asia, poor Caribbean countries). Table 8 offers the mean national income levels in which each party occurs, unweighted by size of party. It is evident that DoEs primarily are a middle-income country phenomenon, whereas pure programmatic and also “Do Nothing” parties predominantly come from rich countries, while clientelistic parties are more typical in very poor countries. Let us also point out that oil

12 and gas rents constitute an exceptionally high share of income in countries with DoE parties. This may certainly add yet another mechanism to funnel resources into political parties that can be deployed for both clientelistic inducements as well as policy expenditures. The gap between Do Everything and Intermediate Mix parties with regards to the state resources available through oil and gas is particularly telling given the previously noted divergence between these two types of parties in their abilities to set up extensive party networks.

Table 8: Average Level of Development at Which Each Party Type Occurs

Intermediate levels of development, however, may supply too thin an explanation for the viability and the success of DoE parties. Any more specific explanation, however, runs the risk that it will seize on sufficiently particular phenomena to become statistically intractable. The main hypothesis, however, is that as a consequence of state-led industrialization there will be plenty of ambitious political elite members who have the networks and resources to contribute to the construction of DoE parties.

There are several pathways to DoEs after state-led development. First, there are politicians who served authoritarian one party regimes and these parties successfully transform themselves into democratic competitors after regime transitions. Examples in our DoEs are from Albania, Macedonia, Moldova, Romania, Serbia, Mexico, Paraguay, Egypt (still an authoritarian hegemonic party when survey took place), Malaysia, and South Korea. Second, there are politicians who were part of the authoritarian ISI/EOI regimes, but after democratic transitions joined new parties in which they could nevertheless deploy their old connections, even though, on the face of it, the new parties were completely separate from the old regime. Cases here would be Albania (once again, with its main non-socialist party), Croatia, Russia, Slovakia, and Mexico (PRD). In a more long-distance fashion, the same logic applies to the parties in advanced industrial democracies that originate in state-led fascist-militaristic development trajectories (Italy, Japan, arguably the Greek Conservatives).

In almost all of these instances, the programmatic appeal of the DoE parties is one that sets them apart from market liberalism. This does not, for the majority of these parties imply a socialist-redistributive appeal, but a milder populist inspiration, or an appeal to security and cross-class alliances to create social order. Market liberalism is seen as a danger that undermines the stability and order achieved in a social fabric that has undergone momentous changes on the way up to middle or high income status.

There is a third track by which politicians and parties derive in the DoE camp, and that is at much lower levels of economic development, on a threshold comparable to that which prevailed in South America or parts of South and East Asia when ISI/EOI strategies took off. These are populist and neo-socialist movements and parties in the poorer regions of Latin America that never went through a deep cycle of ISI development. Party politicians understand that they cannot let their often impoverished constituents’ clientelistic benefits fall by the wayside, but that they also have to project a grand strategic vision of social transformation to get the economy out of the paralysis in which many of these countries found themselves in the 1980s and 1990s. The rise of DoE parties can therefore be observed in poorer Latin American

13 countries—such as Bolivia, Ecuador, Guatemala, and Paraguay, and maybe in the future in Nicaragua or El Salvador.

In addition to an account of DoE party formation that works mainly based on political- economic calculations, add on, as a forth track, a few ethnic parties that represent groups with distinctive positions in the economic division of labor. These parties service their ethnic groups, as groups, with clientelistic benefits, but also call for a programmatic reform of principles of economic distribution with a socialist bent, if the ethnic groups are economically disadvantaged (as in Ecuador, for example), or a market-liberal appeal, if they are advantaged (as in Malaysia).

Tables 9 and 10 provide some evidence as to the empirical plausibility of this story. Table 9 distinguishes levels of lower or higher communist party continuity (based on Kitschelt et al. 1999) and the presence of DoEs in our dataset. Where such continuity is greater, DoEs represent a higher percentage of political parties and also a greater share of electoral support. Yet DoEs exist even in contexts where the old Communist Guard generally disintegrated.

Table 9: Party Linkage Profiles in Post Communist Countries

Table 10 divides up the entire set of DoE parties into four streams. The first stream is neo-socialist parties in poor countries. The second stream combines the different variants of DoE party politics growing out of spells if ISI/EOI development. The third stream lists pure ethnocultural parties that are both programmatic and clientelistic. Finally, there is a residual category of parties and countries that does not fit any of the three arguments. Most parties do fit into one of the first three columns

Table 10. “Do Everything” Parties: Regional Distribution and Electoral Support

While suggestive, the argument is not readily turned into a statistical test. First of all, the numbers of some of the categories are very small. For example, even if parties with strong communist lineage were overrepresented among DoEs the fact that their total may constitute fewer than 10 in a dataset of 506 makes it unlikely that conventional statistical techniques will register this causal stream. Second, we would have to be able to code all of our 506 parties on the theoretically interesting variables. This would not only be an arduous undertaking, in many cases it would also involve a great deal of arbitrary coding decisions that would make the whole statistical exercise questionable.

In light of these obstacles, we see two ways to go. One is resort to comparative case studies to analyze subsets of the DoE countries in greater detail and uncover the causal mechanisms that allow parties to diversify their appeals at the micro level. Also sub national, cross-regional analysis within countries would be immensely valuable in pursuit of that enterprise (cf. Magaloni et al., 2009 on Mexico). The other way is to simplify the causal argument and to subject it to a rough statistical “reality check.” A preliminary version of that undertaking we report next.

Our dependent variable is the type of party strategy that the party uses. We model this choice as a multinomial logit. The first critical independent variable is per capita GDP. If a

14 curvilinear relationship prevails, DoEs should have lower per GDPpc, when compared to pure program parties, but higher GDPpc, when compared to purely clientelistic parties, with no clear predictions offered for parties in the intermediate mix and Do Nothing parties. Our second critical independent variable is oil rents, measured on a per capita base as % of GDP, in order to explore whether an exogenous flow of resources that does not rely on tax extraction from working people in a country facilitates the construction of DoE parties. Our third theoretically interesting independent variable taps ethnic inequalities. It is the measure of income inequality between ethnic groups originally developed by Baldwin and Huber (2010) and now extended to 81 of the 88 countries in the Democratic Accountability dataset by Kolev and Wang (2010). In countries where ethnic inequality is greater, we expect a higher probability of DoE parties, either because specialized ethnic parties take up their cause both through clientelistic benefits to group members as well as a programmatic reversal of the polity’s distributive arrangements or because there are large umbrella parties that venture to contain ethnic disagreements through a partisan- grounded quasi-consociationalism.

The statistical estimation does not include a specific variable to map the ISI-EOI political economic lineage argument we have elaborated above. To a considerable extent, however, the GDP-level related predictor correlates with that variable. Both simple development arguments as well as the political-economy argument predict a curvilinear relationship between affluence and the incidence of DoE parties.

Our model includes three other control variables. First, we introduce the legislative size of parties. Second, we control for quality of democracy (Polity IV score) in order to rule out that DoEs are merely more frequent in regimes that are (still semi-)authoritarian. Finally, we throw in a dummy for Scandinavia, because many Scandinavian countries obtain surprisingly low values of programmatic effort that are hard to explain other than with idiosyncratic anchor points by the Scandinavian experts (see Kitschelt and Freeze 2010).

Because the data are measured at the party and country levels, we need to adjust for heteroskedasticity in the standard errors of the estimates that would otherwise lower the estimated errors of the country-level variables. Standard errors adjusted for clustering by country are sufficient to do so in cases where the country-specific variance is not of theoretical interest (Steenbergen and Jones 2002). The model is estimated using the Do Everything party as the baseline case. However, to facilitate interpretation (given our interest in what explains the emergence of DoE parties instead of what explains their lack of emergence) we present the arithmetic inverse of the estimated coefficient in order to describe the effect of the variable on the likelihood that a party is a Do Everything party. The mathematical transformation of the coefficients is equivalent to estimating 4 multinomial logit models (with each of the other 4 baseline categories sequentially serving as the baseline) and presenting the estimated coefficients for the pairwise prediction between the baseline category and Do Everything parties. While we focus on the predictors of the Do Everything parties, predictors for other party types can be inferred by comparing coefficients from adjacent columns.

Table 11: Predictors of “Do Everything” Strategies

At low levels of development, clientelism predominates. As per capita income increases, the probability of a party choosing the Do Everything, intermediate mix, or Do Nothing strategy instead of the clientelist one all increase. Yet the probability of the pure program strategy is also increasing with income, although the low initial probability of that strategy being used (as manifested by the large positive coefficient for Do Everything strategies when paired with programmatic ones) prevent that strategy from becoming viable until GDP per capita is relatively high. Thus even when controlling for other factors, the Do Everything strategy is most common at intermediate levels of income. The level of development does not, however, significantly differentiate Do Everything parties from parties using a more intermediate mix or from those parties that do not do either clientelism or programmatic issues especially well.

But while development explains the divergence of Do Everything parties from pure program or clientelism, the economic variable that predicts the emergence of Do Everything parties instead of intermediate mixture strategies is the presence of oil and natural gas rents.3 For example, the predicted probability that an average sized party in a country taking the average values on all the variables (and so is thus a middle income country) in the model is a Do Everything party is 0.086 if oil rants are one standard deviation below their mean for the sample. The probability of that party being a Do Everything party doubles to 0.175 if those resource rents increase to a value of one standard deviation above their mean. The change in oil rents lowers the predicted probability of a party using an intermediate mix from 0.43 to 0.28 (with an increase in the probability of pure clientelist competition of 0.05).

Once economic development and resource rents are controlled for, the polity score has no effect on linkage strategy type except in making intermediate mix strategies less common than programmatic ones in very democratic systems. In alternative specifications not presented here we have also tried modeling linkage strategies in accordance with the amount of historical democratic experience a country has, either by testing whether it has been a democracy continuously since 1950 (per Treisman 2007) or using Gerring et al (2005)’s measure of democratic stock experience and none of these variables have a significant impact except for the somewhat curious finding that Do Nothing parties are particularly likely to emerge in older democracies.

The combination of ethnic and economic divisions captured in the BGI/between group income inequality index makes programmatic competition especially unlikely. Relative to pure program parties, DoE parties are much more likely under these circumstances. Yet compared to pure clientelistic parties, DoEs tend to be less common under such circumstances, although the statistical coefficient does not reach any conventional level of significance.

4. Conclusion

Most parties specialize in either a pure programmatic profile, in handing out a large amount of materialist goods, or using a mixture of those strategies that does not seem to do wither one very well. This is consistent with parties facing tradeoffs between the forms that they have to take in order to pursue voters and the kinds of voters they are able to target. Yet some

3 The presence of oil and gas rents becomes significant in differentiating Do Everything parties from the other strategy profiles in models where we control for the use of charismatic, valence, or identity competition. 16 parties break the mold and do everything. In this chapter we have identified some of the clearest examples of these parties. We also have provided a descriptive analysis of these party types:

Pure clientelist parties emerge in poor societies where inequality strongly corresponds to ethnic divisions, tend to be large and may have charismatic leadership, are ideologically centrist and have low levels of party identification to rely on. They also violate campaign finance regulations more than other parties do. Pure program parties are in many ways the opposite of clientelist parties; they emerge in developed countries without strong economic divisions based on class, do better in larger magnitude districts, and take ideological positions outside the center. Do Everything parties are most likely to emerge in middle income countries with large oil reserves. They tend to be large, have extensive party organizations, and centralized control of nominations. They also tend to use other linkage strategy, in particular charisma. Intermediate mix parties are likely to emerge in the same middle income countries as DoE parties but are likely to be smaller, not have a charismatic leader, or access to resource rents. They are also less extensive organizationally than DoE parties are. The analysis provides the least leverage on Do Nothing parties, but suggests that they are common in relatively poor but long standing democracies, are ideologically centrist, and in particular lack a charismatic leader.

The exact causal mechanism underlying these relations remain opaque, however. Either we have not found the elegant simple formula that captures it all, both in sense of spelling out the substantive causal logic as well as in generating a significant statistical effect. Or the causal linkage through authoritarian continuity and legacy are too specific, unfold in a variety of variants, and thus are difficult to code in a straight-forward way, let alone to test statistically. Our goal from this sketch is that the readers will look at Table 2 and Table 10 and see a pattern that we have not.

Works Cited

Baldwin, Kate and John Huber. 2010. Economic versus cultural differences: Forms of ethnic diversity and public goods provision. Forthcoming American Political Science Review.

Cox, Gary W. and Matthew D. McCubbins. 2001. “The institutional determinants of economic policy outcomes.” In Mathew D. McCubbins and Stephan Haggard, eds., Presidents, Parliaments and Policy. New York: Cambridge University Press: 21-63.

Dixit, Avinash and John Londregan. 1996. The determinants of success of special interests in redistributive politics. Journal of Politics 58: 1132-55

Gerring, John, Strom Thacker, and Carola Moreno. 2005. Centripetal Democratic Governance: A Theory and Global Inquiry. American Political Science Review 99: 567-81.

Haggard, Stephan. 1990. Pathways from the Periphery: The Politics of Growth in the Newly Industrializing Countries. Ithaca: Cornell University Press.

Iversen, Torben and Thomas R. Cusack. 2000. The Causes of Welfare State Expansion: Deindustrialization or Globalization? World Politics 52 (3): 313-49.

Keefer, Philip. 2007. Clientelism, Credibility, and the Policy Choices of Young Democracies. American Journal of Political Science 51 (Oct): 804-821.

Kitschelt, Herbert. 2000. Linkages between Citizens and Politicians in Democratic Polities. Comparative Political Studies 33 (6-7): 845-879.

Kitschelt, Herbert and Kent Freeze. 2010. Programmatic Party System Structuration: Developing and Comparing Cross-National and Cross-Party Measures with a New Global Data Set. APSA Annual Meeting, Washington, DC.

Kitschelt, Herbert and Daniel Kselman. 2010. Programmatic party structuration: developing and comparing cross-national and cross-party measures with a new global dataset. APSA Annual Meeting, Washington, DC.

Kitschelt, Herbert, Zdenka Mansfeldova, Radoslav Markowski and Gabor Toka. 1999. Post- Communist Party Systems, Competition, Representation, and Inter-Party Cooperation. New York: Cambridge University Press.

Kolev, Kiril and Yiting Wang. 2010. Ethnic Group Divisions and Clientelism. APSA Annual Meeting, Washington, DC.

Lyne, Mona. 2008. The Voter’s Dilemma and Democratic Accountability: Explaining the Democracy-Development Paradox. College Township: The Pennsylvania State University Press.

Magaloni, Beatriz, Alberto Diaz-Cayeros, and Federico Estévez . 2009. Welfare Benefits, Canvassing and Campaign Handouts. In Jorge Domínguez, Chappell Lawson, and

Alejandro Moreno, Consolidating Mexico’s Democracy. Baltimore: Johns Hopkins University Press.

Shefter, Martin. 1977. Patronage and its Opponents: A Theory and Some European Cases. Ithaca: Cornell University Press.

Singer, Matthew M. 2009. Buying Voters with Dirty Money: The Relationship between Clientelism and Corruption. APSA Annual Meeting, Toronto.

Steenbergen, Marco and Bradford S. Jones. 2002. Modeling Multilevel Data Structures. American Journal of Political Science 46 (Jan): 218-237

Treisman, Daniel. 2007. What Have I Learned About the Causes of Corruption From Ten Years of Cross-National Empirical Research? Annual Review of Political Science. 211-44.

Figure 1: Programmatic and Clientelist Strategies by Party

20 Pearson’s R=-0.4375***

Clientelist Effort

10 5

0 .2 .4 .6 .8 Programmatic Linkage

Figure 2: Linkage Strategy Profiles Program Very Program Very Program Low Program High Low High Clientelism (2) Pure Clientelism (4) Do Everything Very High Clientelism High (3) Intermediate Mix Clientelism of Strategies Low Clientelism (5) Do Nothing Pure Program Very Low

Figure 3: Distribution of Party Strategy Types

Clientelist Effort 5

0 .2 .4 .6 .8 Programatic Linkage

Pure Program Pure Clientelism Intermediate Mix Do Everything Do Nothing

Table 1: Distribution of Party Types by Region

Total Distribution of Parties by Region N Countries with N % of Advanced Post- Latin Asia/Middle One or More Party Parties Africa Parties Industrial Communist America East of This Type Pure Program 126 24.9% 63.7% 23.7% 0.0% 5.1% 9.2% 36 Pure Clientelism 168 33.2% 1.5% 20.3% 52.1% 62.7% 56.1% 48 Intermediate Mix 106 21.0% 10.4% 37.3% 24.0% 5.1% 22.5% 46 Do Everything 44 8.7% 4.4% 13.6% 15.6% 0.0% 7.1% 27 Do Nothing 62 12.3% 20.0% 5.1% 8.3% 27.1% 5.1% 28

Table 2: Strategy Profile Used by the Largest Party (Average Legislative Seats in Last 2 Elections) by Country

Do Everything Clientelistic Programmatic Intermediate None Largest 18 39 13 14 4 party: Albania Angola Australia Belgium Finland enter Bolivia Argentina Austria Botswana New Zealand countries Croatia Bangladesh Canada Brazil Norway in cells Ecuador Benin Czech Rep. Chile Sweden Egypt Bulgaria Denmark El Salvador Estonia Colombia France Ireland Greece Costa Rica Germany Israel Japan Dom. Rep. Latvia Italy Macedonia Georgia Netherlands Lithuania Malaysia Ghana Poland Portugal Mexico Guatemala Switzerland S. Africa Moldova Honduras UK Slovenia Pakistan Hungary USA Spain Romania India Uruguay Russia Indonesia South Korea Jamaica Serbia Kenya Slovakia Lebanon Mali Mauritius Mongolia Morocco Mozambique Namibia Nicaragua Niger Nigeria Panama Paraguay Peru Philippines Senegal Taiwan Tanzania Thailand Turkey Ukraine Venezuela Zambia

Table 3: Electoral Strength by Party Type (Standard Deviation in Parentheses)

Percent Votes Percent Seats Government Status (most recent (legislature in (2008) election before session in 2008) 2008)

Pure Clientelism 17.60 18.91 39.88 (N=168) (17.57) (19.24) (49.11) Do Everything (DoE) 23.98 26.95 61.36 (N=44) (15.74) (17.95) (49.25) Intermediate Mix 12.28 12.93 35.85 (N=106) (12.53) (14.37) (48.18) Pure Programmatic 12.98 13.19 29.37 (N=126) (11.50) (13.87) (45.73) Do Nothing 9.69 10.62 35.81 (N=62) (9.99) (12.53) (44.11)

Table 4: Parties’ Organizational Traits: Extensiveness and Social Networks

Extensiveness of Ties to local Ties to Civil Society Local Offices (a1) Notables Organizations (a3) (a8s) Pure Clientelism 1.98 1.70 2.27 (N=168) (0.70) (0.45) (1.31) Do Everything (DoE) 1.51 1.47 2.09 (N=44) (0.48) (0.32) (1.23) Intermediate Mix 1.81 1.77 1.93 (N=106) (0.55) (0.41) (0.92) Pure Programmatic 1.98 1.94 1.94 (N=126) (0.63) (0.44) (0.79) Do Nothing 2.12 1.99 1.63 (N=62) (0.76) (0.44) (0.88)

Table 5: Clientelistic Effectiveness and Monitoring

Does Not Sanction Effective at Tries to Monitor Voters who Break Clientelism (b11) Voters (c1a) Promise (c4) Pure Clientelism 3.02 0.80 1.56 (N=168) (0.53) (0.19) (0.30) Do Everything (DoE) 3.00 0.84 1.55 (N=44) (0.56) (0.15) (0.28) Intermediate Mix 2.57 0.74 1.77 (N=106) (0.52) (0.17) (0.21) Pure Programmatic 2.19 0.63 1.89 (N=126) (0.44) (0.20) (0.15) Do Nothing 2.24 0.53 1.87 (N=62) (0.49) (0.21) (0.23)

Table 6: Centralization of Party Organization and Party Finance

Local Party Activists Party Following Control Legislative Campaign Finance Nominations Regulations (a5) (a1011e) Pure Clientelism 1.74 0.29 (N=168) (0.49) (0.24) Do Everything (DoE) 1.67 0.54 (N=44) (0.62) (0.38) Intermediate Mix 1.74 0.62 (N=106) (0.50) (0.44) Pure Programmatic 2.15 1.23 (N=126) (0.79) (0.50) Do Nothing 2.05 0.87 (N=62) (0.64) (0.57)

Table 7: Parties’ Efforts with Other than Clientelistic or Programmatic Democratic Accountability Mechanisms

Charisma Party Valence Percent Parties in Identification (Capacity to Govern) Middle Third of (1-4; 4 = High) (1-4; 4 = High) (1-4; 4 = High) Ideological Scale

Pure Clientelism 3.14 2.99 3.15 0.364 (N=168) (0.64) (0.73) (0.52) (0.487) Do Everything (DoE) 3.26 3.26 3.45 0.583 (N=44) (0.55) (0.61) (0.57) (0.494) Intermediate Mix 2.74 2.98 3.09 0.434 (N=106) (0.72) (0.61) (0.62) (0.498) Pure Programmatic 2.60 2.71 2.86 0.364 (N=126) (0.70) (0.65) (0.83) (0.487) Do Nothing 2.43 2.73 2.95 0.468 (N=62) (0.51) (0.55) (0.61) (0.503)

Table 8: Average Level of Development at Which Each Party Type Occurs

Oil/Gas Rents GDP Per Capita Adult Literacy Polity Score as a Percentage (2000 US$) Rate of GDP Pure Clientelism 2,959 80.30 6.94 2.77 (N=168) (3,379) 19.73 (3.52) (7.59) Do Everything 6,267 90.68 7.32 5.22 (DoE) (8,014) (10.09) (2.61) (8.74) (N=44) Intermediate Mix 7,872 92.50 8.30 1.70 (N=106) (8,388) (12.74) (2.84) (5.20) Pure Programmatic 20,923 95.16 8.93 2.06 (N=126) (11,018) (7.66) (2.17) (6.99) Do Nothing 14,638 78.71 7.61 4.67 (N=62) (15,278) (22.26) (3.35) (10.64)

Table 9: Party Linkage Profiles in Post Communist Countries

All Parties Largest Party Lower Higher Lower Higher Total Communist Communist Total Total Communist Communist Total (N Party Party (Percent) (N Parties) Party Party (Percent) Parties) Continuity Continuity Continuity Continuity Pure 37.70% 8.70% 25.23% 27 Program 33.33% 17.65 3 Pure 11.48% 34.78% 21.50% 23 Clientelism 11.11% 50.00% 29.41 5 Intermediate 40.98% 28.26% 35.51% 38 Mix 22.22% 11.76 2 Do 6.56% 19.57% 12.15% 13 Everything 33.33% 50.00% 41.18 7 Do Nothing 3.28% 8.70% 5.61% 6 0.00 0 Total N 61 46 100% 107 Parties 9 8 100 17

Table 10. “Do Everything” Parties: Regional Distribution and Electoral Support

Parties Demanding Parties Growing out of Purely Ethnic Residual Parties State-Led State-Led Economic Parties not accounted Development Development for Bolivia (MaS) Albania: Democratic Party, Socialist Party Poor Ecuador: various Ecuador: Ecuador: Countries Pachakutik various Guatemala: National Macedonia: New Social Macedonia: Unity of Hope Democratic Party IMRO Moldova: Communist Party

Croatia: CDU Colombia: Colombian Conservative Middle Party, Social Income National Unity Countries Party Egypt: NDP Estonia: Centre Party of Estonia Malaysia: UMNO Malaysia: Malay Chinese Association; Mexico: PRI, PRD Mexico: PAN Paraguay: Authentic Rad Lib Party Romania: SDP

Panama: Panameñista Party Russia: United Russia

Serbia: Socialist Party, Serbia: Dem. Radical Party, Party, G17+; New Serbia Slovakia: Direction-Social Democracy; South Korea: Grand Turkey: Natio- National Party nalist Move- ment Party Slovakia: Direction-Social Democracy Greece: New Democracy, Belgium: SP High PASOK

Income Japan: LDP Portugal: SD Countries Italy: Forza Italia

Table 11: Predictors of “Do Everything” Strategies

vs. Pure vs. Pure vs. Intermediate vs. Do Program Clientelism Mix Nothing Legislative Seats 0.068*** 0.014 0.055*** 0.065*** (0.014) (0.010) (0.013) (0.019) Ln(GDP Per Capita) -1.672*** 0.563* -0.299 -0.082 (0.322) (0.265) (0.222) (0.355) Log Oil Rents 0.221 0.172 0.368* 0.205 (0.194) (0.186) (0.169) (0.237) Democracy (Polity Score) 0.109 0.029 -0.068 0.004 (0.062) (0.076) (0.073) (0.078) Income Inequality 38.232*** -19.512 2.621 13.508 Between Ethnic Groups (14.812) (13.862) (13.319) (17.285) Scandinavia -19.945*** 16.854*** 19.524*** -23.780*** (1.306) (1.165) (1.145) (1.205) Constant 2.897* -3.605*** -1.552 -1.660 (1.342) (1.133) (1.042) (1.584) Number of Parties 464 Number of Countries 81 Wald χ2 396.14*** Pseudo R2 0.3808 Multinomial Logit, Standard Errors Clustered by Country in Parentheses * p<0.05, ** p<0.01, *** p<0.001